I think it is a contradiction in terms. I think the Thaler/Sunstein behavioral economics model in "nudge," which made the idea popular, is deeply flawed. It has all the problems of centralized planning (see "Seeing Like a State" by James Scott) dressed up in new language, a more agile trial/error process, and a false sense of security around flawed definitions of rationality.

The main problem is that it is based on a model of rationality that says "rationality=human behavior - biases." It does not recognize that neutralizing biases that make behavior from a model of rational behavior is not enough. You have to justify that model of rationality in the first place against a narrative. The behavioral economists mostly make up laboratory narratives that THEY think is rational (this is the "paternalistic" part: Daddy knows best). That other choices are available makes it "libertarian" apparently. You still have people deciding for others (for example) that eating well is necessary for happiness, and that happiness is actually a good goal that everybody can be assumed to define similarly and seek.

Overall, this is a case of massive overreach: economists and statisticians trying to do philosophy by pretending philosophy doesn't exist or matter.